EUVIMEDEUVIMED entdecken
Crossref

Moisture Interference‐Free Estimation of Soil Salinity Using Proximal Hyperspectral Sensing Through External Parameter Orthogonalization Correction, Feature Extraction and SVM ‐ AdaBoost Modelling

Yiwen Qiu, Jintao Xu, Lin Zhu, Zhiyu Chen, Fangchuan Lou, Shaomin Chen, Shuai Tan

Soil Use and Management · 2026-04 · 42 (2)

DOI: 10.1111/sum.70208 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Soil Use and Management, online erschienen am 16.04.2026. Verfasst von Yiwen Qiu und 6 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 45 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Soil salinity is one of the factors that critically constrains land productivity, necessitating advanced monitoring solutions. This study aims to explore the interference of soil moisture content on the spectral quantitative estimation of soil salinity and to construct a salinity estimation model with moisture interference elimination. A portable spectrometer was used to collect soil proximal spectral information with varying moisture contents and salinity levels based on disturbed soil prepared in the laboratory. The external parameter orthogonalization (EPO) was applied to correct soil spectra, whereas the coupled competitive adaptive reweighted sampling combined with the successive projections algorithm (CARS_SPA) was employed to extract feature wavelengths. Soil electrical conductivity inversion models were established using partial least squares regression (PLSR), random forest (RF), extreme learning machine (ELM) and support vector machine combined with adaptive boosting (SVM‐AdaBoost). Results showed that soil moisture content, electrical conductivity, and their interaction had significant effects on visible and near‐infrared spectral reflectance. The EPO method effectively corrected soil spectra, increasing the number of wavelengths exhibiting a highly significant correlation with soil electrical conductivity ( p < 0.01), and decreasing those significantly correlated ( p < 0.01) with soil moisture. The CARS_SPA extracted 7 feature wavelengths, accounting for only 0.7% of the full spectrum. The SVM‐AdaBoost model performed optimally in salinity retrieval, with R 2 , RMSE and RPIQ of the prediction dataset were 0.861, 333.03 μS·cm −1 and 3.570, respectively. The EPO‐CARS_SPA‐SVM‐AdaBoost framework effectively eliminates moisture interference in salinity inversion. This methodology provides a theoretical basis and reference for estimating soil parameters.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Qiu, Y., Xu, J., Zhu, L., Chen, Z., Lou, F., Chen, S., & Tan, S. (2026). Moisture Interference‐Free Estimation of Soil Salinity Using Proximal Hyperspectral Sensing Through External Parameter Orthogonalization Correction, Feature Extraction and SVM ‐ AdaBoost Modelling. Soil Use and Management, 42(2). https://doi.org/10.1111/sum.70208

Vancouver

Qiu Y, Xu J, Zhu L, Chen Z, Lou F, Chen S, et al. Moisture Interference‐Free Estimation of Soil Salinity Using Proximal Hyperspectral Sensing Through External Parameter Orthogonalization Correction, Feature Extraction and SVM ‐ AdaBoost Modelling. Soil Use and Management. 2026;42(2). doi:10.1111/sum.70208

BibTeX
@article{Qiu2026,
  author = {Qiu, Yiwen and Xu, Jintao and Zhu, Lin and Chen, Zhiyu and Lou, Fangchuan and Chen, Shaomin and Tan, Shuai},
  title = {Moisture Interference‐Free Estimation of Soil Salinity Using Proximal Hyperspectral Sensing Through External Parameter Orthogonalization Correction, Feature Extraction and SVM ‐ AdaBoost Modelling},
  journal = {Soil Use and Management},
  year = {2026},
  volume = {42},
  number = {2},
  doi = {10.1111/sum.70208}
}
Kostenlos starten →